Anthropic pledges 5% of its shares to a giving fund but hasn't said who will run it

Anthropic, the company behind the Claude chatbot, plans to put 5% of its shares into a new entity called the Anthropic Public Benefit Fund. The pledge, announced on October 18, still needs approval from the company's stockholders. After that, Anthropic's board can add another 2% each time the company's value rises 50% above a baseline it hasn't disclosed.

The fund could make large grants to nonprofits or send money directly to specific groups of people, such as workers whose jobs AI makes less secure. Who picks the recipients, under what rules and when spending starts are all open questions. Anthropic says details on structure and governance will come "in the coming months."

What 5% of Anthropic is worth

Anthropic didn't put a dollar figure on the fund. Its most recent public valuation comes from a May 28, 2026 funding round, when it raised $65 billion at a $965 billion post-money valuation, TechCrunch reported. At that price, a 5% stake would be worth about $48 billion.

That figure is an estimate based on a private round, and the fund won't be holding $48 billion in cash. It will own shares in a company that isn't publicly traded yet, although Anthropic confidentially filed draft IPO paperwork with the SEC in June. The stake turns into money only when someone buys those shares, and its value will rise or fall with Anthropic's.

Anthropic says the new fund, its co-founders' pledge to give away 80% of their equity and a program that matches stock donated by employees together set aside roughly 20% of the company's shares for the public good.

Why a company-owned fund and not a foundation

Anthropic is a public benefit corporation registered in Delaware. It's a for-profit company with investors who expect a return, but its charter also names a public purpose: developing advanced AI responsibly for the long-term benefit of humanity. As Anthropic explained in 2023, that legal form lets its board weigh the mission against stockholders' financial interests.

A foundation or charity has no owners collecting profits and must spend its assets on its mission. In the US, charities that take tax-deductible donations usually register under section 501(c)(3) of the tax code. Anthropic skipped that route and set up the fund as a limited liability company it owns outright, which it says lets it mix different kinds of giving. That also means the fund isn't independent of the company. Anthropic did set one limit: shares held by the fund won't be used for lobbying or political donations.

Jobs come first on the list

The example of workers hurt by AI ties back to an earlier pledge. In July, Anthropic put $200 million into its Economic Futures Research Fund, which pays universities, research institutes and nonprofits to study how society can prepare for AI's economic effects, including new forms of income support for people who lose work to automation.

That fund pays for research. The new one could pay affected people directly, though Anthropic hasn't said how or to whom. The question gets harder to dodge as AI models move from answering questions to attempting entire jobs.

Why now

Anthropic's only stated reason is that the pace of AI progress may call for "bolder approaches." TechCrunch ties the move to growing public hostility toward AI. In August, after an investor accused CEO Dario Amodei of feeding that backlash with his warnings, Amodei called it "fundamentally a crisis of trust" and conceded that AI companies, his own included, haven't yet delivered on their big promises, TechCrunch reported.

The same weekend, Politico published a draft AI safety bill Anthropic has been pitching to US lawmakers, according to PC Gamer. The bill, which a spokesperson said reflects proposals Anthropic has pushed since June, would let the government act only when a model poses a foreseeable, imminent risk of more than $1 billion in damage or of killing or seriously injuring more than 50 people. Anthropic has lobbied in Washington before, joining OpenAI against Chinese open-weight models.

Who decides, and what we still don't know

The only decision the announcement assigns so far, adding each extra 2%, belongs to Anthropic's board. That board doesn't answer only to investors. Since 2023, an independent body Anthropic created, the Long-Term Benefit Trust, has elected a growing share of its members.

OpenAI set things up the other way around. After its October 2025 restructuring, the nonprofit OpenAI Foundation kept equity then valued at about $130 billion and controls the company, according to OpenAI. At Anthropic, the company owns the fund.

Several answers are still missing: who will manage the fund, how recipients will be chosen, what baseline triggers each extra 2%, and whether the fund will publish its finances.